UnboundLocalError & non-iterate problems

Nobody has claimed this yet.

Hello, I think your program is perfect, and the visualizing results are so great.
However, I encountered some problems when I implement your program.

Running the cantilever.py may cause an error just like this:

UnboundLocalError: local variable 'output_phi' referenced before assignment.
-------------------------------------------------------------------------------
which is in letop/optimization/interface.py according to the error above.
Well, if I set output_phi=0 (any value) at first. There won't occur this error.

But it won't iterate anymore. It just shows:

Warning, merit function did not decrease (merit=55.37948162421951, new_merit=55.37948162421951)-> Trial 1
Time stepping of Hamilton Jacobi failed. Reason: 0
Linear firedrake_1_ solve converged due to CONVERGED_ITS iteration 1
new_J=15.587277790787676, newC=[2.e-05]
merit=55.37948162421951, new_merit=55.37948162421951
Warning, merit function did not decrease (merit=55.37948162421951, new_merit=55.37948162421951)-> Trial 2
Time stepping of Hamilton Jacobi failed. Reason: 0
Linear firedrake_1_ solve converged due to CONVERGED_ITS iteration 1
new_J=15.587277790787676, newC=[2.e-05]
merit=55.37948162421951, new_merit=55.37948162421951
.........
Warning, merit function did not decrease (merit=55.37948162421951, new_merit=55.37948162421951)-> Trial 10
Time stepping of Hamilton Jacobi failed. Reason: 0
Linear firedrake_1_ solve converged due to CONVERGED_ITS iteration 1
new_J=15.587277790787676, newC=[2.e-05]
merit=55.37948162421951, new_merit=55.37948162421951

I don't know how to solve the problem of not iterating, maybe setting the output_phi=0 is wrong at the beginning.
I’d be grateful for your reply.

Start with cantilever.py and the output_phi control flow in letop/optimization/interface.py; reproduce the UnboundLocalError before changing initialization. Then investigate why the reported merit remains unchanged and Hamilton-Jacobi stepping fails, and verify that the example completes iterations without the warnings.

Written by the indexing model from the issue text.